Chris@16
|
1 /*
|
Chris@16
|
2 *
|
Chris@16
|
3 * Copyright (c) 1998-2002
|
Chris@16
|
4 * John Maddock
|
Chris@16
|
5 *
|
Chris@16
|
6 * Use, modification and distribution are subject to the
|
Chris@16
|
7 * Boost Software License, Version 1.0. (See accompanying file
|
Chris@16
|
8 * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
|
Chris@16
|
9 *
|
Chris@16
|
10 */
|
Chris@16
|
11
|
Chris@16
|
12 /*
|
Chris@16
|
13 * LOCATION: see http://www.boost.org for most recent version.
|
Chris@16
|
14 * FILE regex_split.hpp
|
Chris@16
|
15 * VERSION see <boost/version.hpp>
|
Chris@16
|
16 * DESCRIPTION: Implements regex_split and associated functions.
|
Chris@16
|
17 * Note this is an internal header file included
|
Chris@16
|
18 * by regex.hpp, do not include on its own.
|
Chris@16
|
19 */
|
Chris@16
|
20
|
Chris@16
|
21 #ifndef BOOST_REGEX_SPLIT_HPP
|
Chris@16
|
22 #define BOOST_REGEX_SPLIT_HPP
|
Chris@16
|
23
|
Chris@16
|
24 namespace boost{
|
Chris@16
|
25
|
Chris@16
|
26 #ifdef BOOST_MSVC
|
Chris@16
|
27 #pragma warning(push)
|
Chris@16
|
28 #pragma warning(disable: 4103)
|
Chris@16
|
29 #endif
|
Chris@16
|
30 #ifdef BOOST_HAS_ABI_HEADERS
|
Chris@16
|
31 # include BOOST_ABI_PREFIX
|
Chris@16
|
32 #endif
|
Chris@16
|
33 #ifdef BOOST_MSVC
|
Chris@16
|
34 #pragma warning(pop)
|
Chris@16
|
35 #endif
|
Chris@16
|
36
|
Chris@16
|
37 #ifdef BOOST_MSVC
|
Chris@16
|
38 # pragma warning(push)
|
Chris@16
|
39 # pragma warning(disable: 4800)
|
Chris@16
|
40 #endif
|
Chris@16
|
41
|
Chris@16
|
42 namespace re_detail{
|
Chris@16
|
43
|
Chris@16
|
44 template <class charT>
|
Chris@16
|
45 const basic_regex<charT>& get_default_expression(charT)
|
Chris@16
|
46 {
|
Chris@16
|
47 static const charT expression_text[4] = { '\\', 's', '+', '\00', };
|
Chris@16
|
48 static const basic_regex<charT> e(expression_text);
|
Chris@16
|
49 return e;
|
Chris@16
|
50 }
|
Chris@16
|
51
|
Chris@16
|
52 template <class OutputIterator, class charT, class Traits1, class Alloc1>
|
Chris@16
|
53 class split_pred
|
Chris@16
|
54 {
|
Chris@16
|
55 typedef std::basic_string<charT, Traits1, Alloc1> string_type;
|
Chris@16
|
56 typedef typename string_type::const_iterator iterator_type;
|
Chris@16
|
57 iterator_type* p_last;
|
Chris@16
|
58 OutputIterator* p_out;
|
Chris@16
|
59 std::size_t* p_max;
|
Chris@16
|
60 std::size_t initial_max;
|
Chris@16
|
61 public:
|
Chris@16
|
62 split_pred(iterator_type* a, OutputIterator* b, std::size_t* c)
|
Chris@16
|
63 : p_last(a), p_out(b), p_max(c), initial_max(*c) {}
|
Chris@16
|
64
|
Chris@16
|
65 bool operator()(const match_results<iterator_type>& what);
|
Chris@16
|
66 };
|
Chris@16
|
67
|
Chris@16
|
68 template <class OutputIterator, class charT, class Traits1, class Alloc1>
|
Chris@16
|
69 bool split_pred<OutputIterator, charT, Traits1, Alloc1>::operator()
|
Chris@16
|
70 (const match_results<iterator_type>& what)
|
Chris@16
|
71 {
|
Chris@16
|
72 *p_last = what[0].second;
|
Chris@16
|
73 if(what.size() > 1)
|
Chris@16
|
74 {
|
Chris@16
|
75 // output sub-expressions only:
|
Chris@16
|
76 for(unsigned i = 1; i < what.size(); ++i)
|
Chris@16
|
77 {
|
Chris@16
|
78 *(*p_out) = what.str(i);
|
Chris@16
|
79 ++(*p_out);
|
Chris@16
|
80 if(0 == --*p_max) return false;
|
Chris@16
|
81 }
|
Chris@16
|
82 return *p_max != 0;
|
Chris@16
|
83 }
|
Chris@16
|
84 else
|
Chris@16
|
85 {
|
Chris@16
|
86 // output $` only if it's not-null or not at the start of the input:
|
Chris@16
|
87 const sub_match<iterator_type>& sub = what[-1];
|
Chris@16
|
88 if((sub.first != sub.second) || (*p_max != initial_max))
|
Chris@16
|
89 {
|
Chris@16
|
90 *(*p_out) = sub.str();
|
Chris@16
|
91 ++(*p_out);
|
Chris@16
|
92 return --*p_max;
|
Chris@16
|
93 }
|
Chris@16
|
94 }
|
Chris@16
|
95 //
|
Chris@16
|
96 // initial null, do nothing:
|
Chris@16
|
97 return true;
|
Chris@16
|
98 }
|
Chris@16
|
99
|
Chris@16
|
100 } // namespace re_detail
|
Chris@16
|
101
|
Chris@16
|
102 template <class OutputIterator, class charT, class Traits1, class Alloc1, class Traits2>
|
Chris@16
|
103 std::size_t regex_split(OutputIterator out,
|
Chris@16
|
104 std::basic_string<charT, Traits1, Alloc1>& s,
|
Chris@16
|
105 const basic_regex<charT, Traits2>& e,
|
Chris@16
|
106 match_flag_type flags,
|
Chris@16
|
107 std::size_t max_split)
|
Chris@16
|
108 {
|
Chris@16
|
109 typedef typename std::basic_string<charT, Traits1, Alloc1>::const_iterator ci_t;
|
Chris@16
|
110 //typedef typename match_results<ci_t>::allocator_type match_allocator;
|
Chris@16
|
111 ci_t last = s.begin();
|
Chris@16
|
112 std::size_t init_size = max_split;
|
Chris@16
|
113 re_detail::split_pred<OutputIterator, charT, Traits1, Alloc1> pred(&last, &out, &max_split);
|
Chris@16
|
114 ci_t i, j;
|
Chris@16
|
115 i = s.begin();
|
Chris@16
|
116 j = s.end();
|
Chris@16
|
117 regex_grep(pred, i, j, e, flags);
|
Chris@16
|
118 //
|
Chris@16
|
119 // if there is still input left, do a final push as long as max_split
|
Chris@16
|
120 // is not exhausted, and we're not splitting sub-expressions rather
|
Chris@16
|
121 // than whitespace:
|
Chris@101
|
122 if(max_split && (last != s.end()) && (e.mark_count() == 0))
|
Chris@16
|
123 {
|
Chris@16
|
124 *out = std::basic_string<charT, Traits1, Alloc1>((ci_t)last, (ci_t)s.end());
|
Chris@16
|
125 ++out;
|
Chris@16
|
126 last = s.end();
|
Chris@16
|
127 --max_split;
|
Chris@16
|
128 }
|
Chris@16
|
129 //
|
Chris@16
|
130 // delete from the string everything that has been processed so far:
|
Chris@16
|
131 s.erase(0, last - s.begin());
|
Chris@16
|
132 //
|
Chris@16
|
133 // return the number of new records pushed:
|
Chris@16
|
134 return init_size - max_split;
|
Chris@16
|
135 }
|
Chris@16
|
136
|
Chris@16
|
137 template <class OutputIterator, class charT, class Traits1, class Alloc1, class Traits2>
|
Chris@16
|
138 inline std::size_t regex_split(OutputIterator out,
|
Chris@16
|
139 std::basic_string<charT, Traits1, Alloc1>& s,
|
Chris@16
|
140 const basic_regex<charT, Traits2>& e,
|
Chris@16
|
141 match_flag_type flags = match_default)
|
Chris@16
|
142 {
|
Chris@16
|
143 return regex_split(out, s, e, flags, UINT_MAX);
|
Chris@16
|
144 }
|
Chris@16
|
145
|
Chris@16
|
146 template <class OutputIterator, class charT, class Traits1, class Alloc1>
|
Chris@16
|
147 inline std::size_t regex_split(OutputIterator out,
|
Chris@16
|
148 std::basic_string<charT, Traits1, Alloc1>& s)
|
Chris@16
|
149 {
|
Chris@16
|
150 return regex_split(out, s, re_detail::get_default_expression(charT(0)), match_default, UINT_MAX);
|
Chris@16
|
151 }
|
Chris@16
|
152
|
Chris@16
|
153 #ifdef BOOST_MSVC
|
Chris@16
|
154 # pragma warning(pop)
|
Chris@16
|
155 #endif
|
Chris@16
|
156
|
Chris@16
|
157 #ifdef BOOST_MSVC
|
Chris@16
|
158 #pragma warning(push)
|
Chris@16
|
159 #pragma warning(disable: 4103)
|
Chris@16
|
160 #endif
|
Chris@16
|
161 #ifdef BOOST_HAS_ABI_HEADERS
|
Chris@16
|
162 # include BOOST_ABI_SUFFIX
|
Chris@16
|
163 #endif
|
Chris@16
|
164 #ifdef BOOST_MSVC
|
Chris@16
|
165 #pragma warning(pop)
|
Chris@16
|
166 #endif
|
Chris@16
|
167
|
Chris@16
|
168 } // namespace boost
|
Chris@16
|
169
|
Chris@16
|
170 #endif
|
Chris@16
|
171
|
Chris@16
|
172
|